Fully Functional Logic‐In‐Memory Operations Based on a Reconfigurable Finite‐State Machine Using a Single Memristor

Abstract

resistance switching characteristic have been extended from the traditional nonvolatile memory [3] to the new computing paradigms. [4][5][6][7][8][9] A notable field is the memristor-based logic-in-memory (LIM) concept, which is capable of constructing an in-memory computation system by merging the function of the arithmetic logic unit (ALU) and memory, which are separated in the conventional von Neumann architecture. [10] Figure 1a illustrates the typical configuration of the conventional von Neumann computing machine. The machine is divided into three parts: the main memory, control unit, and ALU. The four-step machine cycles of fetching, decoding, executing, and storing are used to control the instruction and data streams for the performance of a computing process. The disparity of operation speed between the central processing unit (CPU, ALU + control unit) and the memory incurs operation delay, and the huge data transfer between the memory and the ALU, especially in data-intense tasks, consumes too much energy. This problem is called "von Neumann bottleneck," which is regarded as the fundamental limitations of the current computers. [11,12] This bottleneck can be addressed by performing computation steps within the memory block, which is called "in-memory computing (IMC)" or "logic-in-memory (LIM)." Figure 1b shows the schematic structure of such computational machines. When the ALU and memory are merged into a compact memory-ALU (MALU), the results can be saved while the commands are being executed. In this way, the storing step is omitted, leading to a highly energy-efficient computing paradigm. The memristor can provide fine-grained functionality support for the IMC machine because the logic output can be saved along with the logic operation when the LIM concept is adopted. This important feature may render the ideal IMC machine feasible. Figure 1c shows an approximate possible layout of the IMC machine, a kind of hybrid complementary metal-oxide-semiconductor (CMOS)/memristor crossbar architecture. The CMOS layer is used as the control unit whereas the memristor crossbar array layer and a small-amount CMOS auxiliary layer constitute MALU for LIM operations. This structure shares some of the features of the neuromorphic computation architecture based on memristors, [6] but it is not a network-based structure. Apparently, the memristor-based LIM Memristor offers a promising logic-in-memory (LIM) functionality to achieve the futuristic in-memory computing machine, which may solve the problem of the "von Neumann bottleneck" in the conventional computer architecture. A sequential logic concept is capable of achieving LIM based on the finitestate machine (FSM) using a single bipolar (BRS) or complementary resistive switching (CRS) memristor, where 14 of the 16 two-input Boolean logic functions (XOR and XNOR are missing) are mapped between the resistance state and the terminal voltages. In this paper, a new FSM-LIM concept based on a reconfigurable finite-state machine (RFSM logic) is proposed, which is experim...

Resistive Memory‐Based In‐Memory Computing: From Device and Large‐Scale Integration System Perspectives

“…According to the available stable resistance states, RRAM can be categorized into two types: analog RRAM denotes those devices whose resistances can be programmed to any value between the highest resistance state (HRS) and the lowest resistance state (LRS), whereas binary RRAM behaves as a normal memory device with a stable HRS and LRS. [48][49][50][51][52][53][54] Binary inputs are stored into RRAM devices as the resistance (conductance) before performing computation. As early as in 2009, a 1 kb RRAM array in a one-transistor-one-RRAM (1T1R) cell structure was successfully integrated with CMOS read/write circuits.…”

Recent Advances in Resistive Switching Materials and Devices: From Memories to Memristors

“…With this broadened definition, it is immediately realized that resistive switching memories are particular digital cases of the memristors that exhibit binary conductance states. Characterized by the continuous and incremental change of device conductance or resistance, memristors with analog-type switching behaviors have recently triggered the research campaign of the novel in-memory computing paradigm including neuromorphic computation and logic 11,[170][171][172][173][174][175][176][177][178][179][180][181][182][183][184] applications.…”

A Stateful Logic Family Based on a New Logic Primitive Circuit Composed of Two Antiparallel Bipolar Memristors

“…One of them can be regarded as analog‐type in‐memory computing in which the vector–matrix multiplication is used as the primitive to perform the computation . The other corresponds to logic in memory (LIM), in which the computation primitives are the basic logic operations (or gating) . The former seems to attract more extensive attention because performing the vector–matrix multiplication by physical means (crossbar array [CBA]) based on Kirchhoff's law is a great replacement for digital operations using the conventional CMOS circuit for machine learning and deep learning .…”

“…The logic operation is achieved by mapping the logic variables of input and output to the circuit variables of the resistance, voltage, or current of the memristor. According to the symmetry of the logic variables, the current memristor‐based LIM strategies can be subdivided into two categories of “memristor sequential logic” and “stateful logic.” The logic inputs of voltage and logic outputs of resistance (or current) are the characteristics of the memristor sequential logic. In this case, one of the research goals is achieving different (as many as possible) logic operations in only one memristor, utilizing the underlying mathematical relationships .…”